Re: svn: /php/php-src/ branches/PHP_5_3/ext/exif/exif.c branches/PHP_5_3/ext/exif/tests/bug54002.phpt branches/PHP_5_3/ext/exif/tests/bug54002_1.jpeg
branches/PHP_5_3/ext/exif/tests/bug54002_2.jpeg trunk/ext/exif/exif.c trunk/ext/exif/tests

why don't you fix it the way you want instead of arguing endlessly? I gave all info to explain the reasoning behind that code, if it is wrong, and that's totally possible, then fix it and let move on :) Cheers, On Tue, Feb 15, 2011 at 2:21 PM, Dmitry Stogov <dmitry@zend.com> wrote: > On 02/15/2011 03:31 PM, Pierre Joye wrote: >> >> On Tue, Feb 15, 2011 at 12:54 PM, Dmitry Stogov<dmitry@zend.com>  wrote: >> >>> I don't know all the logic but it looks very strange >>> >>> 2147483648 == 0x80000000 != 0x7fffffff == 2147483647 >>> >>> so byte_count_signed may be 0x80000000 == -1 >> >> negative value are checked before. Btw, byte_count_signed  is a 64bit >> signed integer, for this reason :) > > Lets byte_count_signed = 0x0000000080000000 > > it's not less than zero and not greater than 2147483648. > So the following condition is false. > > if (byte_count_signed < 0 || (byte_count_signed > 2147483648)) { > > Then you convert it to size_t. > > byte_count = (size_t)byte_count_signed; > > In case size_t is defined as signed int32 you will get byte_count = -1. -- Pierre @pierrejoye | http://blog.thepimp.net | http://www.libgd.org